Planning, Inference and Pragmatics in Sequential Language Games

TL;DR

This paper introduces a comprehensive model for sequential language games that incorporates inference, pragmatic reasoning, and goal-oriented communication, validated on a new crowdsourced dataset to better understand human-like interactions.

Contribution

The paper presents a novel model integrating inference, pragmatics, and goal-oriented messaging in sequential language games, advancing understanding of human communication strategies.

Findings

Model effectively captures human behavior in language games

Pragmatic reasoning improves communication success

New dataset enables better evaluation of language game models

Abstract

We study sequential language games in which two players, each with private information, communicate to achieve a common goal. In such games, a successful player must (i) infer the partner's private information from the partner's messages, (ii) generate messages that are most likely to help with the goal, and (iii) reason pragmatically about the partner's strategy. We propose a model that captures all three characteristics and demonstrate their importance in capturing human behavior on a new goal-oriented dataset we collected using crowdsourcing.